Vals AI found that teams of AI agents barely beat single agents while costing up to 5.1 times more, with only one of four tests using GPT-6 Sol and Claude Opus 5.5 showing a measurable gain. Anthropic's own data shows quality plateaus beyond ten agents while token costs keep rising.
In the US, personal AI agents comb through bank statements for forgotten or duplicate subscriptions and then talk directly to customer service to lower bills. Insurers and telecom operators that profited from customer inertia now face a tireless counterpart.
A developer describes using MCP to give Claude Code context about a system they built, so that new sessions do not start from scratch. The agent located the app's shared contract in two calls, but the experiment hit a limit once the available records were exhausted.
Epoch AI and Anthropic separately found that models such as GPT-5.6 Sol and Claude Fable 5 can run experiments but lack scientific self-criticism and genuine creativity. Sol reached at best 15% of the human reference score, using methods researchers already knew.
Traditional enterprise systems assume jobs finish quickly, retries are free and identical inputs give identical outputs. Agentic workloads break all three, which is why successful pilots can become operational headaches once they run unattended.
AI agents are now used widely for coding, customer service, IT operations and complex enterprise tasks. A SiliconANGLE column argues they are nonetheless not the right fit for every organization.